Figures 3 a, b and c A within (PC1=68.17%, PC2=28.04%), (b)
between (PC1 99.61%, PC2= 0.31%) and (c) Multivariate analysis of
variation (discriminant analysis) sampling locations Principal component
analysis with convex hull at 95% ellipses showing the contributions to
variations observed from the three locations (Red= Kwara), Blue and
Green, Ibadan and Makurdi sample populations respectively
Size and shape variations in samples from Ibadan in Fig. 3a were more
disposed on the first axis with size contributing a higher proportion of
variations observed in comparison to those from Kwara and Makurdi where
shape variation was an overriding factor as they were better disposed on
the second axis. Between the geographical location groups (Fig. 3b),
Makurdi samples demonstrated the highest population variation in both
size and shape factors compared to samples from other geo-locations.
Large area of overlaps existed between samples evaluated.
Analysis of similarities (ANOSIM) at 9999 permutations demonstrated
significant (P˂0.1) Mahalanobis distance of 0.045 in Kwara samples and
0.910 in Ibadan sample population. Multivariate analysis of variance
(MANOVA) (Fig. 3c) failed to reveal any significant (P˂0.7) discriminant
values based on geography. Wilks lambda (0.936), df=8, F=0.5994